J. Physiol. (1970), 206, pp. 437-455 437 With 6 text-ftgure8 Printed in Great Britain CONSEQUENCES OF MONOCULAR DEPRIVATION ON VISUAL BEHAVIOUR IN KITTENS BY P. B. DEWS AND T. N. WIESEL From the Laboratory of Psychobiology, Department of Psychiatry, and the Department of Neurobiology, Harvard Medical School, Boston, Massachusetts, 02115, U.S.A. (Received 11 August 1969) SUMMARY 1. Kittens were raised for a period with one of their eyes closed by suture of the lids. The age at suture and the duration of deprivation were varied systematically. When the cat was a year or more old, the normal and deprived eyes were compared using behavioural procedures which made graded demands on visual function. 2. In kittens deprived from birth, the duration of eye closure determined the severity of the defect in vision with the deprived eye. A cat with an eye closed for the first 4-6 weeks showed as a permanent effect only a lowering of the visual acuity. When closure was extended through the first 7 weeks the visual acuity was further lowered but the animal still showed good visual guidance of paw placement. Further extension of deprivation through the first 16 weeks led to a still more severe defect; such animals showed no indication of visual guidance of paw placement or of pattern discrimination. They were influenced visually by stimuli that differed in luminosity. 3. The upper age limit of the susceptibility to deprivation was determined by varying the age at eye closure. Waiting until 1 month of age before closing the eye conferred no appreciable protection. Waiting until 2 months of age, however, reduced the damage. Deprivation starting at 4 months of age or later produced no effect we could detect. Thus, susceptibility is greatest during the second month after birth and then falls until by 4 months of age the kitten, like the adult cat, suffers no permanent consequences of monocular light and form deprivation. 4. After exclusive use of the deprived eye for a period, brought about by closure of the normal eye, visual control with the deprived eye was better than in similarly deprived cats whose normal eye was never closed. Improvement in the deprived eye was also seen in an animal whose normal eye was closed after both eyes had been open for more than one year. I15-2
438 P. B. DEWS AND T. N. WIESEL 5. Relating the behavioural results to the neurophysiological findings in the visual cortex in the same or similarly deprived cats shows that the grading of visual defects with age and length of deprivation was generally paralleled by a change in proportion of cortical cells driven by stimulation of the deprived eye. The effect of reversal of eye closure in improving behavioural control was not, however, accompanied by an increase in the ability of the deprived eye to drive cortical cells. INTRODUCTION Clinical observations in children with congenital cataracts have shown that lack of normal visual experience of form during early life causes severe defects in vision (von Senden, 1960). Similar effects have been demonstrated in behavioural studies of animals raised in darkness or under diffuse light conditions (see review by Riesen, 1966). Studies of the physiological and morphological effects of raising kittens with one eye closed by eyelid suture during the first few months after birth revealed changes in both the lateral geniculate and the visual cortex (Wiesel'& Hubel, 1963a, b). Simple behavioural observations indicated that these animals also suffered gross and persistent defects of vision with the deprived eye. In the present study, behavioural tests have been adapted to assess objectively and more precisely the defects produced in kittens by monocular deprivation. The dependency of the visual defects on the timing and duration of deprivation has been determined and so have the effects of closing the normal eye when the deprived eye was opened. Some of the kittens in the series and others with a similar history of deprivation were studied in acute neurophysiological experiments (Hubel & Wiesel, 1970), permitting direct correlation between cortical physiology and the control of behaviour by visual stimuli (visual stimulus control). METHODS Apparatus The experiments were carried out in a plywood runway 180 cm long and 60 cm wide (Fig. 1 A). Two translucent plastic panels in the front wall were transilluminated as visual stimuli. The panels were hinged at the top. Behind each panel was a hole to which a food cup could be raised. The cats pushed back the panels with their noses, bringing their mouths over the holes for the food cups. At the other end (the back) of the runway there was a hemispherical knob in the floor and a hole for a food cup similar to those behind the panels. Cats shuttled from back to front of the runway pressing first the knob and then one of the panels and then the knob and so on until a series of trials was completed. Pressing an illuminated panel extinguished the panel lights and lit the knob; pressing the lighted knob darkened the knob and lit the panels. Food was presented for 3 sec when the positive panel was pushed but not when the negative panel was pushed. Food was
BEHAVIOUR AFTER MONOCULAR DEPRIVATION 439 delivered when the illuminated knob was pushed following a positive panel push but not following a negative panel push. In other words, food was delivered at both ends of the runway following presses of the positive panel but at neither end following presses of the negative panel. Images were projected on the panels by a pair of Kodak Carousel Projectors. The image was restricted to an 11-5 cm high and 10 cm wide part of the panel as shown in Fig. 2. Preliminary observations were made with 1, 2 and 3 log-unit neutral density filters taped in front of the projection lens. With the 2 log-unit filters the black lines of the stimuli (see below) showed against a white background of about +2 log cd m-2, a reasonable and effective level of illumination, which was then used in all the reported experiments. A photocell beside the panel was activated by the positive stimulus, programming food delivery when the panel was pressed. A 0 10 20 30 40 50 cm Right panel Front Food cupeo Back Partition ~~~Knob 0 Left panel B Front platform Back platform 10 20 30 cm 0~~~~~~~~~~~~~~ _Z4= Right panel 10 = Front Icood cup Left panel 40 ~~~~~~Knob" --_-Water Back Fig. 1. A. Diagram of runway. The plastic panels were mounted in chassis on the front wall and transilluminated by the projectors. The open circles indicate the position of the holes in the floor to which the food cups could be raised. The filled circle near the back of the runway shows the knob, which was positioned so that the cat could readily put its right foot on the knob and its mouth over the hole for the food cup. The dashed lines in the runway show the position of the partition in the Acuity Test. B. Arrangement for Water Barrier Test. The dish of water, the front and back platforms and the pier have been added to the runway shown in A. Between series of trials, the distance from the back platform to the end of the pier could be changed by sliding the front of the dish of water under the front platform, keeping the back platform at the back of the dish. Between trials a handle attached to the pier was used to move the pier sideways to a new position. The handle ran horizontally under the front platform to the outside of the runway.
440 P. B. DEWS AND T. N. WIESEL The Gap Test The positive stimuli were four continuous horizontal black lines across the lighted stimulus area (Fig. 2). The lines were 0-5 cm wide and were separated by spaces of 1-5 cm. The negative stimuli were similar except there was a gap in the lines. Gaps about either 30 mm or 1-5 mm wide were used. An unilluminated panel was another negative stimulus. Since the cat pushed the panel with its nose, it was not possible, in the ordinary Gap Test, to specify the effective visual angle subtended by the gaps. Some observations were made with a modification of the Gap Test, referred to as an Acuity Test. An opaque centre partition of adjustable length was inserted extending Fig. 2. Cat's eye view of front wall of runway. The outline of the two panels and the areas where the stimuli appeared on the panels are shown and the position of the centre partition is indicated. The stimuli are those for the Gap Test, with the positive stimulus on the left. back from the front wall of the runway (Fig. 1A). An error was scored if the cat approached the negative stimulus panel closer than the end of the partition. The visual angle subtended at the end of the partition by the gap in a line was calculated. In a Luminescent Panel Test the plastic panels were lit by electroluminescent panels (General Electric, Type FN) whose luminance depended on the voltage applied. The positive panel had a luminance of + 0 7 log cd m-2, the negative panel of -15 log cd m-2. The Water Barrier Test A square metal dish was put across the middle of the runway and filled with water to a depth of about 3 cm (Fig. IB). The sides of the dish were vertical and fitted flush with the sides of the runway. Platforms 15 cm above the floor of the runway
BEHAVIOUR AFTER MONOCULAR DEPRIVATION 441 were arranged in front of the dish and behind it. From the front platform a pier of wood 5 cm wide projected perpendicularly backwards for 44 cm, some of that distance being over the water. As in the Gap Test, the cat shuttled between the back and the front of the runway, passing each time now over the water by means of the pier and platforms. The position of the pier was changed laterally to a new position each time the cat retumed to the back of the runway. The distance from the back platform to the end of the pier could also be varied from 10 to 30 cm. When the distance was 10 cm, cats could stand on the rear platform and reach the pier with outstretched paw. When the distance was 30 cm, the pier could be reached only after the centre of gravity of the cat had passed forward over the water under the guidance of only visual cues. In experiments with the water barrier only the positive panel was transilluminated in each trial. The room was dimly lit by reflected light, providing ambient illumination in the mesopic range. The times required for the cat to move from knob to stimulus panel were measured. Programme In each projector were ten slides, five positive and five negative, arranged so that there were no more than two consecutive slides of the same kind (positive or negative). A positive slide in the right-hand projector was always accompanied by a negative slide in the left-hand projector and vice versa. The two projectors' slide holders moved together in phase from 1 to 10 and then back to 1 and so on. Slides were changed only when a panel had been pushed but not after every panel push. Whether a change would occur or not was determined by a three-position switch, ABC (programmed by an Automatic Electric Series OCS Relay), which changed after each push of the positive panel and which cycled indefinitely, ABCABC...A push of the positive panel led to a slide change when the relay was in the A or C position but no change when it was in the B position. A push of the negative panel led to a slide change when the relay was in the C position but no change when it was in the A or B positions. Since pushing the negative panel did not change the OCS relay position, a negative stimulus on the panel on a particular side could not, two times out of three, be changed to a positive stimulus by repeatedly pushing that side following each knob press. The positive stimulus was equally often on the right or on the left. Which side would be positive in a given trial, however, was not independent of what had happened in the previous trial because of the constraints of the sequence of slides and of the OCS programme. When the positive panel was pushed, there was a 14/27 probability that the positive panel would be on the opposite side on the next trial. When the negative panel was pushed, there was a probability of 7/27 that the positive panel would be on the opposite side on the next trial. With consecutive pressings of the negative panel the probability was either 0 that the positive panel would change sides (on the two out of three occasions that the OCS was in position A or B), thus discouraging consistenitly going to the same side, or 7/9 that the positive panel would change sides (on the one out of three occasions that the OCS was in position C), thus discouraging alternation. Criteria of Btimulu8 control If the cat was equally likely to go to the right or to the left, but independently of the stimuli, the probability was 0-5 that the response on a given trial would be on the positive stimulus. From the binomial distribution, the expected number of correct responses by chance in a series of 30 trials would be 15 with a standard deviation of 2-7. Fewer than 10 errors would be expected by chance, therefore, in less than 1
442 P. B. DEWS AND T. N. WIESEL series in 20 (i.e. P < 0.05). For the Gap Test, the scores in three consecutive series of 30 trials have been pooled; a score of 30 errors or fewer in 90 trials has a probability of occurring by chance of less than 0-001, and has been taken as evidence of stimulus control. Subject8 There were fifteen cats in the main experimental series, divided into three groups depending on their history of visual deprivation. Cats in Group I had their right eyes closed by lid suture (Wiesel & Hubel, 1963 a) 8-15 days after birth, before they had had any significant visual experience, and then opened at various ages to determine how long initial deprivation must be continued to produce a defect (Table 1). Cats in Group II had their right eyes closed at various ages to determine to what age stusceptibility to the effects of deprivation persisted (Table 2). Finally, cats in Group III had their right eyes closed for varying periods starting 8-15 days after birth, as for Group I, but then the left eyes were closed when the right eyes were opened, to determine whether exclusive use of the previously deprived eye would enhance its functional capabilities (Table 3). One additional cat was studied, for immediate transfer of stimulus control. After lid separation all cats had clear corneas, normal appearing fundi, and good direct and consensual pupillary light reflexes. Procedure The cats were partially deprived of food, trained to eat promptly from the food cups and then trained to press the panels and the knob. Individual eyes were tested by occluding the other eye with an opaque contact lens. At first only the positive panel was lighted. Meanwhile the cats became accustomed to working with a contact lens in one eye. The cats were then trained under the Gap Test. At first, the positive stimulus was moved slightly up and down, a manoeuvre which seemed to help the establishment of stimulus control. Subsequent performances under the Water Barrier Test, the Acuity Test and the Luminescent Panel Test required essentially no additional training. Results were accepted as definitive only when the cat had been trained several months and was at least a year old. Following initial training, body weights could be returned to the normal range, without detriment to performance in the various tests. RESULTS With their normal left eyes, all the eleven cats of Groups I and II went predominantly to the positive stimulus with the continuous lines rather than the negative stimulus with 30 mm gaps in the lines (Fig. 2). They made an average of 4-1 errors in 90 trials. Nine of these cats were studied with the 1-5 mm gap stimuli and, again, all were controlled by the stimuli, averaging 8-9 errors in 90 trials with no cat making as many as 20 errors. The same nine cats studied in the Water Barrier Test completed series of trials with no trial time in excess of 30 sec, and median times less than 10 sec (Table 1). These results with normal eyes provide a standard for comparison with the results with deprived eyes. One great advantage of monocular closure is that conclusions can be based on differences between the normal and deprived eyes of the same individual.
BEHA VIOUR AFTER MONOCULAR DEPRI VATION 443 Group I cats, with early monocular clo8ure8 of graded duration Lids, sutured shortly after birth, were opened in two cats at about 4 months of age (Cats 3 and 10) and in one cat at about 5 months (Cat 2). The resulting deficiency in vision is shown by the complete lack of control with the deprived right eye by even the 30 mm gap (Table 1). Testing was continued for as long as 2 years without noticeable improvement. TABLE 1. Deprivation history and behavioural performances of cats in Group I Tables 1-3 give the ages in days since birth when eyes were closed and then opened. L and R refer to performances with the normal left eye and with the deprived right eye only, the other eye being occluded by an opaque contact lens. The number of errors that occurred in three consecutive series of 30 trials are given for steadystate performance in the Gap Test. For the Water Barrier Test the medians of times in sec for the animal to go from knob to stimulus panel over 30 cm of water in 10 consecutive trials are given. The infinity sign means the cat did not cross the water Performance Deprivation Water -A Gap Barrier Right eye Test Test (days) (errors) (sec) Cat r, -, no. Litter Closed Opened mm L R L R 2 A 8 145 30 7 38 3 B 15 111 30 18 31-10 C 15 i11 30 0 1-5 7 35 8-4 oc 19 D 8 51 30 2 11 1-5 4 37 5-4 488 12 E 10 37 30 0 2 1-5 3 22 6-8 6-5 13 E 10 37 30 2 4 1.5 7 23 4-0 4-7 14 F 10 32 30 4 7 1.5 4 29 4.4 4 0 The deficiency was dramatically demonstrated in the Water Barrier Test with Cat 10. (Cats 2 and 3 were dead before this test and the 1-5 mm Gap Test were introduced.) With the normal left eye open, Cat 10, like all the cats studied in the test, moved from back to front of the runway, over the water, without hesitation, completing a trial in only a few seconds (Table 1) and then moved with equal facility from front to back of the runway, again over the water, to start a new trial. With an opaque contact lens in
444 P. B. DEWS AND T. N. WIESEL the normal left eye and the pier at a distance of 10 cm, Cat 10 approached the water barrier slowly, over a period of more than 30 sec, then groped for the pier by extending a paw with abducted claws, in dramatic contrast to the elegant precision with which the paw was placed on the pier when the normal left eye was unoccluded. When the extended paw touched the pier, the cat passed over the water rapidly; and two subsequent trials with the pier in the same position were completed in progressively shorter times (though they were still more than 30 sec). On the next trial the pier 15 _ ~ Cat 19 10 5 0 @1~~F- Ḋ0 En 10 z fillcat 20 5 SI t1 30 mm gap 1-5 mm gap Fig. 3. Performance in Gap Test. Performance in three consecutive series of 30 trials is shown for the 30 mm gap and the 1-5 mm gap. Note the consistency in performance in consecutive series of trials. Cat 19 made many more errors with only its deprived right eye unoccluded than with its normal left eye unoccluded. Litter-mate Cat 20, with similar initial deprivation but subsequent normal left eye closure, performed almost as well with only its deprived right eye as with its normal left eye. OI left eye; 0 right eye. was moved; the cat stepped forward, put its foot where the pier had been and landed with both front paws in the water. The cat retreated and did not cross the water barrier again that session. In subsequent sessions, with the pier at 30 cm, the cat regularly gave satisfactory performances with the normal left eye uncovered but never completed the test with only the deprived right eye uncovered.
BEHAVIOUR AFTER MONOCULAR DEPRIVATION 445 In training, Cats 2, 3 and 10, with their normal left eyes occluded, went readily to the lighted rather than the darkened panel, proving some persistent visual function through the right eye. In the Luminescent Panel Test, Cat 10 made 14/90 errors with both eyes open and 12/90 errors with only the deprived right eye open; that is, Cat 10 performed as well with only its deprived eye unoccluded as with its normal eye open, so there was no evidence that deprivation impaired control by the luminescent panels. 10 Cat 12 5 0-10 Catl3 E z 30 mm gap 1-5 mm gap Fig. 4. Performance in Gap Tests. Arrangement as in Fig. 3. Note again the consistency in performance in consecutive series of trials and also the virtual identity of performance with the corresponding eyes of the littermates similarly deprived. The control exerted by the 1-5 mm gaps with the deprived eyes was statistically highly significant (P < 0-001); yet the difference between the control exerted by the 30 mm gap and the 1 5 mmr gap, and the difference between the control with the two eyes stand out clearly. When the right eye had been closed for shorter periods, cats were controlled not only by the light and dark panels but also by other stimuli in the tests. Cat 19, deprived until 7 weeks of age, came under significant control in the 30 mm Gap Test with its deprived right eye (14/90). Control was better, however, through the undeprived eye (7/90 errors) (Table 1). The difference between the two eyes was much greater in the 1-5 mm Gap Test in which Cat 19 was not controlled with its deprived right eye. The results in three consecutive series of 30 trials are shown in Fig. 3, illustrating the consistency of performance. In the Water Barrier Test, the animal performed satisfactorily with only the deprived right eye open.
446 P. B. DEWS AND T. N. WIESEL Thus deprivation for 7 weeks produced a severe permanent defect, but a significantly less severe defect than 4 months of deprivation. After a still shorter period of eye closure, 4- to 52 weeks (Cats 12, 13 and 14), stimulus control with the deprived eye was no longer limited to the Water Barrier Test and the 30 mm gaps but now extended to the 1*5 mm gaps. The control by the 1-5 mm gaps, however, although significant with the deprived eye, was better with the normal eye (Fig. 4). The difference 20 Cat 12 15 - : ~~~~~~~~~~~~Cat 13 10 i Partition length (cm) Fig. 5. Performance in Acuity Test. Errors in series of 30 trials with centre partitions of different lengths are shown for Cats 12 and 13. Significant control (less than 10 errors) was exerted with the normal eye3 with partition lengths up to 50 cm, but the control with the deprived eye was lost at a distance of only 15 cm, confirming the difference between the eyes suggested by the results in Fig. 4. Note again the virtual identity of performance with the corresponding eyes of litter-mates similarly treated. O left eye, right E eye. between the eyes was accentuated in the Acuity Test on Cats 12 and 13. With their normal left eyes open both cats made fewer than 10 errors with partition lengths up to about 50 cm, where a 1*5 mm gap subtends a visual angle of about 10 min of arc. With their deprived right eyes both cats made more than 10 errors with a partition length of only 15 cm (visual angle 33 man of arc) (Fig. 5). In summary, deprivation until only 42 weeks of age produced a defect
BEHAVIOUR AFTER MONOCULAR DEPRIVATION 447 which persisted more or less unchanged over the succeeding months. The defect was progressively greater with increasing duration of deprivation until after more than 3 months there was no evidence of visual function except by light and dark panels. Group II cats with later monocular closure To obtain information on the beginning and end of the period of susceptibility to visual damage by eye closure, the eyes of the cats in Group II were closed when the kittens were already 1, 2 or 4 months of age. TABLE 2. Cat no. 15 21 16 18 Deprivation history and behavioural performances of cats in Grouip II Deprivation A Right eye (days) Performance Gap Test (errors) Litter Closed Opened mm L R F 32 122 30 5 44 1.5 12 G 30 47 30 3 13 1-5 11 24 F 63 147 30 4 4 1.5 16 28 F 114 199 30 0 0 1.5 16 11 Water Barrier Test (sec) L R 2'2 o* 2-6 3-0 4-2 4.5 3-3 3-0 * When the normal left eye of Cat 15 was later closed, the cat completed the Water Barrier Test (see Discussion). Closure from 4- weeks of age until 4 months (Cat 15) produced as severe a defect as closure from the early days of life until 4 or 5 months of age. Cat 15 failed to come under control of the 30 mm gaps and did not perform the Water Barrier Test (Table 2). When closure was reduced to the period from 4 weeks to 61 weeks of age (Cat 21) the animal performed well on the Water Barrier Test and came under the control of both the 30 mm and 1-5 mm gaps (Table 2). With the 1.5 mm gaps, however, 24/90 errors were made, which is outside the range of errors of cats with their normal eye (3 to 18) and is very similar to performance of Cats 12 and 13 whose eyes were closed until 51 weeks of age. Thus the period of susceptibility continues beyond the first month of life, and the severity of the defect is related to the duration of closure. When closure was delayed until the period from 2 months of age until 5 months (Cat 16), the subsequent defect was much less than that observed
448 P. B. DEWS AND T. N. WIESEL in animals deprived for a similar period starting earlier in life (Cats 2, 3, 10 and 15); Cat 16 not only completed the Water Barrier Test but also came under the control of the 30 mm gaps and, partially, the 1b5 mm gaps. Thus, although it persists beyond 2 months of age, susceptibility is already declining by that age. Closure from 4 until 7 months of age (Cat 18) produced no detectable defect, showing that susceptibility to damage from closure is lost before a kitten is 4 months old. In an additional cat, Cat 17, the right eye was closed at 6 months of age (on the 172nd day of life) and the cat was trained in the runway while the right eye was closed. After 3 months of closure (on the 262nd day) the right eye was opened under halothane anaesthesia. About 3 hr later, when the animal had recovered from the anaesthesia, the left eye was occluded and performance in the Gap Test assessed. From the outset, performance was satisfactory, 1 error in the first 30 trials with the 1V5 mm gap. Not only was there no detectable impairment of function of an eye closed 3 months when closure was delayed until 6 months of age, but no training specifically with the deprived eye was necessary for it to exert control in situations learned while it was closed. Group III cats, with right eye closed early, then left eye closed when right eye opened The left eyes of the cats in Groups I and II were never closed by lid suture. If a normal eye continues to suppress a deprived eye even after opening of the lids, the suppression would be removed by closing the normal eye. Closure of the normal eye might also induce some recovery in the deprived right eye that otherwise might not have occurred. To investigate these possibilities, the four cats of Group III had their left eye closed when the right eye was opened (Table 3). The first two cats (Cats 8 and 9) were litter-mates of Cat 10 and all three were subjected initially to similar closures of the right eye until about 4 months of age. As a result of the subsequent closure of the normal left eyes, Cats 8 and 9 came to perform satisfactorily in the Water Barrier Test over the next few months while Cat 10, which did not have reversal of eye closure, continued to fail dismally with its deprived right eye. There may be a hint of impairment in Cats 8 and 9 in the Water Barrier Test in that the times were longer with the deprived right eye than when the normal left eye was available; nevertheless, the performance of these cats was decisively better than that of litter-mate Cat 10. Neither Cat 8 nor Cat 9, however, came under the control of even the 30 mm gap, showing a persistent severe visual defect. Reversal of closure improved the visual function of kittens even when
BEHA VIOUR AFTER MONOCULAR DEPRI VATION 449 duration of deprivation was shorter and the initial defect was correspondingly less severe. While Cat 19, deprived for 7 weeks, failed to come under the control of the 1-5 mm gap, litter-mate Cat 20 with similar initial deprivation but subsequent reversal made only 20/90 errors, demonstrating highly significant control, though errors were still above the normal range (Table 3). Figure 3 illustrates the improvement produced by eye reversal by comparing Cats 19 and 20 in the 30 mm and 1-5 mm Gap Tests. It should be remembered that the left eye of Cat 20 was closed for only 3 months and that the tests were done when both eyes had been open for some months. TABLE 3. Deprivation history and behavioural performances of cats in Group III Performance Deprivation, A A Water Gap Barrier Right eye Left eye Test Test (days) (days) (errors) (sec) Cat, _, no. Litter Closed Opened Closed Opened mm L R I, R 9 C 15 111 111 244 30 1-5 6 17 42 4-6 10-6 8 C 15 i11 i11 679* 30 1.5 2 3 38 1*2 4*8 1 H 9 94 94 465 30 1*5 8 24 8 18 12-6 15*0 20 D 8 52 52 252 30 1-5 4 15 4 20 1*1 1.3 * The left eye of Cat 8 was opened for testing between the 244th and 343rd days of age. There was no obvious difference in visual function between Cats 8 and 9, so the second period of closure in Cat 8 had little effect. Cat 1, with the right eye closed for the first 3 months, then the left eye closed, represents a special case for two reasons. First, performance level in the Water Barrier Test, and, to a lesser extent in the Gap Test, was below that of other cats in the series. These derelictions were almost certainly due to a mild though unmistakable ataxia in the hind limbs that also caused the cat to fall off the pier occasionally during the WVater Barrier Test. None of the other animals was so clumsy. Secondly, testing of Cat 1 continued over a very long period, almost 5 years. By the end of 44 years' testing Cat 1 was under the control of not only the 30 mm gaps but also the 1-5 mm gaps, representing definitely better performance than Cats 8 and 9 with similar deprivation history, but shorter subsequent testing. There was no evidence of impairment of function with the left eyes in the cats of Group III in spite of prolonged closure and even though, in the
450 P. B. DEWS AND T. N. WIESEL case of Cat 20, the eye was closed as early as 7 weeks. Cat 20 confirms that the effects of deprivation are greatly reduced by the time the kitten is 2 months old. The comparison of Cat 16 (deprived from 2 to 5 months of age) and Cat 20 suggests that susceptibility is still further reduced when vision with the other eye has already been damaged by deprivation. This suggestion brings to mind the neurophysiological finding that monocular closure produces more damage to vision with the deprived eye than does binocular closure (Wiesel & Hubel, 1965a). A normally functioning eye seems to contribute to the damage to the other closed eye. TABLE 4. Comparison of behavioural and neurophysiological information on visually deprived cats Cat Visual number defectt in in Deprivation behavioural deprived Percentage of cellst history* series eye driven by deprived eye 4 months 10 + + + + 21 (9) from Fig. 9A (same cat) 7 weeks 19 + + 20 (10) from Fig. 8B 54 weeks 12 + 26 (25) from Fig. 7A (same cat) 54 weeks 13 + 39 (38) from Fig. 7B (same cat) 3 months starting at 16 + 32 (32) from Fig. 5B 2 months of age 3 months starting at 18 0 90 (90) from Fig. 6A (same cat) 4 months of age 4 months starting at 17 0 80 (80) from Fig. 6B 6 months of age 3 months then closure of 1 + 11 (7) from Fig. 9B (same cat) normal eye * Starting within 2 weeks of birth unless noted otherwise. t From Fig. 6. Each test in which a cat failed to come under contiol was scored one + for summarizing the degree of defect. t From Hubel & Wiesel (1970); numbers derived from Figures of that paper, as indicated. Cells refers to neurones in the visual cortex. The figures in parentheses give the percentage of cells showing normal selectivity among stimuli with different orientations and movements. DISCUSSION The striking morphological and neurophysiological changes in the visual system of kittens deprived monocularly in early life led us to this study of the effects of deprivation on some behavioural activities depending on vision. Our results, summarized graphically in Fig. 6, led to the following conclusions on the relations between neurophysiological and behavioural consequences of deprivation (Table 4). First, permanent defects were produced only when an eye was closed during a strictly limited period of susceptibility. Normal kittens do not open their eyes until 1-2 weeks after
BEHAVIOUR AFTER MONOCULAR DEPRIVATION 451 birth, so visual experience is not necessary during that period. Neurophysiological studies indicate that the period of susceptibility begins around the early part of the fourth week. This suggestion accords with our finding that the behavioural defect appeared identical following closure to about 4 months of age beginning either at 1 month of age (Cat 15) or shortly after birth (Cats 2, 3 and 10). The period of susceptibility ends before a kittenl is 4 months old; after that age closure produces neither Period of closure (days) 0 100 200 Behavioural results A 1-5 30 W Group I 2 1 2 L 3 1// Group II 10 I 19 12 E 13 El 14L3 15y/// 21 H 161 18EL Group III 9 //////A L V/i// zzz2& 8 1 Ez2z2m I amm 20,,, I I I I I I I _ I I I _ I II _I I I II L I I -I Fig. 6. Graphical summary of results. The deprivation history is shown on the left and the behavioural results on the right. A, Acuity Test; 1-5, 1-5 mm Gap Test; 30, 30 mm Gap Test; W, Water BarTier Test. 'Closure' indicates the period the eye was closed (0). 'Control' means significant control of the behaviour of the cat by the stimuli of the test (C]). 'No control' means: for the Water Barrier Test, failure to complete the test; for the Gap Tests, more than 30 errors in 90 consecutive trials; for the Acuity Test, more than 10 errors in 30 trials (-). For each cat, the upper row shows the results for the right eye and the lower row for the left eye. The purpose of the Figure is to provide a general view of the results; for details see Figs. 1, 2 and 3. Cats 2 and 3 were studied experimentally only in the 30 mm Gap Test.
452 P. B. DEWS AND T. N. WIESEL behavioural nor neurophysiological consequences. Secondly, even within the period, susceptibility was not constant. A 2j weeks' closure during the second month (Cat 21) had an effect equivalent to a 3 months' closure beginning at the end of the second month (Cat 16). Again the neurophysiological results are in good agreement, showing that cells in the visual cortex are more resistant to effects of monocular deprivation after the second month. Thirdly, the severity of the permanent defect was graded, depending on the number of days of deprivation during the second and third months after birth, from mere loss of acuity (Cats 12, 13 and 14) to complete loss of all visual stimulus control except to stimuli that differed in luminosity (Cats 2, 3, 10 and 15). Concordantly it was shown neurophysiologically that during the period of susceptibility, even a few days' deprivation caused marked changes, and that increased duration of deprivation increased the defect (Hubel & Wiesel, 1970). Further, more cortical cells could be driven normally in cats deprived 51 weeks than in cats deprived 3-4 months (compare Cats 12 and 13 with Cat 10, Table 4). Fourth and lastly, closure of the normal eye when the initially closed right eye was opened led to an improvement in the visual control of behaviour with the deprived eye. This improvement was not, however, accompanied by any substantial change in the ability of the deprived eye to influence cells in the visual cortex in an acute physiological experiment (compare Cats 1 and 10, Table 4; see also Wiesel & Hubel, 1965b). Stimulus control, the constraint of behaviour by discriminative stimuli, varies in degree, as is illustrated by the finding that the normal eye always gave better visual control than did the deprived eye (Fig. 6). Neurophysiological results have been that more cortical neurones can be driven with a normal eye than with a deprived eye. Taken together, these results suggest that the number of cortical cells influenced by an eye determines its power to control vision-dependent behaviour. The limiting condition is reached after 3 or 4 months' deprivation when behaviour is influenced visually only by light and dark panels and when only a small proportion of cortical neurones are driven normally with the deprived eye (about 10 % ). The slowly increasing behavioural control with the deprived eye that results from closing the normal eye is most simply ascribed to the persisting cortical neurones driven by the deprived eye being able to acquire control when not chronically masked by the signal from the normal eye; but the possibility that the recovery is mediated through a less damaged non-cortical component of the visual system cannot be excluded. In any event we found that once the deprived eye had acquired control it was not lost when the normal eye was reopened. Indeed, the behavioural control with the deprived eye could continue to improve after the normal eye was reopened, as in Cat 1. When the deprivation effect was initially
BEHA VIO UR AFTER MONOCULAR DEPRI VATION 453 less, closing the normal eye could improve the control of the deprived eye even to the verge of the normal eye range (compare Cats 19 and 20). If the improvement in behavioural control exerted by the deprived eye when the normal eye is closed is based mainly on those neurones that persisted in their responsiveness to the deprived eye through the period of closure, it might be expected that closure of the normal eye at any time in life would result in improved visual control with the deprived eye. Preliminary observations have been made in Cat 15, which showed control only by the contrast between the lighted and unlighted panels with its deprived eye up to more than a year of age. When the normal eye was then closed, within a few weeks the cat was crossing the Water Barrier for the first time in its life with only its deprived eye open. It appears, then, that the processes involved in recovery following reversal of eye closure are entirely different in nature from the processes maintaining functional integrity under normal circumstances through the sensitive period of the first months of life; behavioural recovery resulting from reversal can probably take plaoe at any time in the life of the cat and is not accompanied by parallel changes in responsivity of cortical neurones. With their normal eyes, cats were controlled by a stimulus attribute, the 1-5 mm gap in the lines, subtending a visual angle of about 10 min of arc. The object of the tests was to identify differences between normal and deprived eyes, and not to explore systematically to find the optimum stimulus configuration, level of illumination and viewing distance for maximum visual acuity. Nevertheless, the figure of 10 min of arc agrees reasonably with the 5-5 min of arc reported by Smith (1936) for a comparable '2-point' type of stimulus. Single line stimuli could exert control even when they subtended visual angles of less than 2 min of arc (Smith, 1936), a finding which agrees with assessment based on optokinetic responses (Ganz & Fitch, 1968). Smith (1936) remarks that 'visual acuity' as measured by single lines is approximately five times more precise than when determined by the separation of two lines. Several authors have considered the possibility of defects in visual control of movement, distinct from loss of acuity, following visual deprivation (Riesen, Kurke & Mellinger, 1953; Meyers & McCleary, 1964; see review by Held, 1967). The Water Barrier Test required visual control of paw placement but made only modest demands on visual acuity. The test was a semi-automated development of Hein & Held's prong test (Hein & Held, 1967). The differences in control with normal and deprived eyes were consistently less in the Water Barrier Test than in the Gap Test. Failures in the Water Barrier Test occurred only when the loss of acuity was gross enough to account for the incapacity. In addition, even cats which failed the Gap and Water Barrier Tests had no difficulty in guiding themselves
454 P. B. DEWS AND T. N. WIESEL toward the brighter of two lighted panels. Thus our results give no evidence of a persisting defect in visuo-motor control followiilg deprivation. Transient defects in visual guidance as described by Hein & Held (1967) are clearly different in nature from the permanent defects in acuity found in our cats. The partial recovery from the effects of deprivation should not be allowed to obscure the essentially permanent nature of changes in the visual system when deprivation occurs during the period of susceptibility. The effects of 2 weeks deprivation were easily detectable behaviourally many months later and would probably continue for the life of the cat. Even after as short a period of deprivation as 3 days, the visual system shows anatomical and physiological changes (Hubel & Wiesel, 1970) that may have permanent sequelae. The findings suggest that, clinically, every effort should be made to avoid the interference with normal visual experience in early life which would result from covering an eye for treatment of infection or other pathology. Finally, the studies on eye reversal suggest that some recovery in visual function after treatment of older children may not signify improvement in the primary visual pathways up to and including the striate cortex, but only the better exploitation of visual information already influencing cortical neurones. This research was supported by Grant NB 0554-01 from the U.S. Public Health Service. We are indebted to Mrs Marilyn Vanderhoof Bownds for help with the experiments and to Miss Eleanor Bates for help with the manuscript. REFERENCES GANZ, L. & FITCH, M. (1968). The effect of visual deprivation on perceptual behavior. Expl Neurol. 22, 638-660. HEIN, A. & HELD, R. (1967). Dissociation of the visual placing response into elicited and guided components. Science, N.Y. 158, 390-392. HELD, R. (1967). Dissociation of visual functions by deprivation and rearrangement. Psychol. Forsch. 31, 338-348. HUBEL, D. H. & WIESEL, T. N. (1970). The period of susceptibility to the physiological effects of unilateral eye closure in kittens J. Physiol. 206, 419-436. MEYERS, B. & MCCLEARY, R. A. (1964). Interocular transfer of a pattern discrimination in pattern deprived cats. J. comp. physiol. Psychol. 57, 16-2 1. RIESEN, A. H. (1966). Sensory deprivation. In Progress in Physiological Psychology, vol. 1, pp. 117-147, ed. STELLAR, E. & SPRAGUE, J. M. New York: Academic Press. RIESEN, A. H., KURKE, M. I. & MELLINGER, J. C. (1953). Initerocular transfer of habits learned monocularly in visually naive and visually experienced cats. J. comp. physiol. Psychol. 46, 166-172. SMITH, K. U. (1936). Visual discrimination in the cat: IV. The visual acuity of the cat in relation to stimulus distance. J. genet. Psychol. 49, 297-313.
BEHAVIOUR AFTER MONOCULAR DEPRIVATION 455 VON SENDEN, M. (1960). Space andl Sight: The Perception of Space and Shape in the Congenitally Blinid Before anl After Operation. Glencoe, Ill.: The Free Press. WIESEL, T. N. & HUBEL, D. H. (1963 a). Effects of visual deprivation on morphology and physiology of cells in the cat's lateral geniculate body. J. Neurophysiol. 26, 978-993. M IESEL, T. N. & HUBEL, D. H. (1963b). Single-cell responises in striate cortex of kittens deprived of visioln in one eye. J. Neutrophysiol. 26, 1003-1017. MWIESEL, T. N. & HUBEL, D. H. (19653a). Comparisoni of the effects of unilateral anid bilateral eye closure on cortical uinit responses in kittens. J. Neurophysiol. 28, 1029-1040. WVIESEL, T. N. & HUJTBEL, D. H. (1965b). Extent of recovery from the effects of visual deprivation in kittenis. J. Neurophysiol. 28, 1060-1072.